HERPETOLOGICAL JOURNAL, Vol. 9, pp. 163-167 (1999)

FEEDING HABITS OF SYMPATRIC DJSCOGLOSSUS MONTALENTII, DISCOGLOSSUS SARDUS AND EUPROCTUS MONTANUS DURING THE BREEDING SEASON

SEBASTIANO SALYIDI01 , ROBERTO SINDAC02 AND LlVIO EMANUELl3

'Istituto di Zoologia, Universita di Genova, Via Balbi 5, I- I6126 Genova, Italy

2Istituto per le Piante da Legno e Ambiente, Corso Casale 476, 1- 10I 32 Torino, Italy

1Acquario di Genova, Area Porto Antico - Ponte Sp inola, I- 16128 Genova Italy

The diets of three Corsican amphibians, Discoglossus montalentii, Discoglossus sardus and Euproctus montanus, were studied in the Ospedale region during the breeding season. Adult specimens were collected in or around breeding pools and were stomach flushed in the field. Prey taxa included a large variety of terrestrial and aquatic prey items of variable size, indicating opportunistic predation. All species were able to catch their prey both on land and in water, but varied in the proportions of aquatic and terrestrial prey consumed. E. montanus fed largely upon benthic macroinvertebrates, suggesting predation in deep water; D. sardus mainly captured terrestrial prey; and D. montalentii showed a mixed fe eding strategy, preying upon both terrestrial and aquatic prey categories in similar proportions. Discoglossus sardus showed the highest standardized value of niche breadth (D, = 0. 769), compared to D. montalentii and E. montanus (D, = 0.544 and D, = 0.523 respectively). When prey size frequency distributions were compared, no specific differences were observed. These results indicated that, at least during the breeding season, trophic segregation among sympatric amphibians was maintained by different foraging strategies, and that the three species exploited contiguous microhabitats in different ways.

Key words: Corsica, Discoglossus, Euproctus, foragingstrategy, sympatry

INTRODUCTION STUDY AREA AND METHODS On the Mediterranean island of Corsica seven am­ Samples were collected in the Ospedale forests phibian species, including three endemic forms (Southern Corsica, 20 km NW of Portovecchio) from (Euproctus montanus, Discoglossus montalentii and April to June, both in 1996 and 1997. The study area, a Salamandra corsica), are known to occurr (Delaugerre Pinus laricio forest ranging from 700 to 1000 m a.s.l., & Cheylan, 1992; Gase et al., 1997). The Corsican was located between the villages of Ospedale and painted frogD. montalentii was the last species to have Zonza, which are separated by a linear distance of been described on the basis of biochemical and mor­ about 12 km. Although very similar morphologically, phological characters (Lanza et al., 1984 ), and -due to D. montalentii and D. sardus were identifiedin the fi eld its confusion with the congeneric Tyrrhenian painted mainly on the basis of their hind limb length, which is frog D. sardus - little is known about its biology and relatively greater in D. montalentii (Clarke & Lanza, ecology. Discoglossus montalentii is mainly a moun­ 1990; Salvidio et al., unpublished data); in some cases, tain species living in pristine or semi-pristine forest fieldid entification was confirmed by dissection and by streams, while D. sardus breeds from sea level to about chromosomal analysis (Gaetano Odierna in litteris). 1300 m, in a variety of aquatic habitat types (Clark & With the exception of Salamandra corsica (see be­ Lanza, 1990; Lanza, Vanni & Brizzi, 1992a, b). The low), all amphibians were captured in or near (i.e. less altitudinal distribution of the two species overlaps than I metre from) water. All specimens were meas­ broadly, and in several areas they are found together ured from thetip of the snout to the posterior end of the (Clarke & Lanza, 1990; Salvidio et al., unpublished vent (SVL) to the nearest mm, and toe clipped at first data). During a two year field study on the distribution capture; thus, stomach contents were analyzed only and ecology of D. montalentii and D. sardus, we col­ once for each individual. Diet composition was ob­ lected data on their fe eding habits together with those tained by stomach flushing (Fraser, 1976; Joly, 1987), of the sympatric Corsican brook salamander E. and stomach contents were preserved in the field in montanus. This gave us the opportunity to analyse their 70% ethanol. Stomach flushing was performed at least feeding behaviours and to compare their foraging twice on each specimen and, if no prey was obtained, modes in order to investigate whether they were parti­ the stomach was considered empty. Where possible, tioning food resources when they coexist. food items were identified to order or family, and meas­ ured to the nearest 0.1 mm under a dissecting Correspondence: S. Salvidio, Istituto di Zoologia, Universita microscope. Aquatic invertebrates were classified us­ di Genova, Via Balbi 5, 1-16126 Genova, Italy. E-mail: [email protected] ing Sansoni's (1988) identification keys of freshwater 164 S. SAL VIDIO, R. SINDACO AND L. EMANUELI

macroinvertebrates. Prey volume (V)was calculated by Table 2 gives the complete data set of the diet com­ considering each item as a cylinder or a sphere, and was position by taxonomic group and by volume. In the expressed in mm3• Additionally, prey items were classi­ Ospedale forests, amphibians consumed a wide range fied as aquatic or terrestrial, with a third category of terrestrial and aquatic invertebrates. The Corsican comprising prey that dwell on land, in shallow water or brook salamander was the only species that preyed mud (e.g. metamorphosing anurans, mites, dipteran lar­ upon vertebrates: one unidentified urodelan larva and vae, earthworms), as well as unidentified preyitems. one postrnetamorphic froglet (Discoglossus sp.) were Simple correspondence analysis (CA), based on a found in the stomachs of two males, and a third male contingency table with predators as rows and prey taxa swallowed four eggs belonging to its own species. The as columns, was used to obtain a graphical representa­ distribution of prey items according to their life-style tion of the association between species and prey indicated that all species were able to feed both in categories. In this analysis, a small sample of the Corsi­ aquatic and terrestrial environments; however, E. can fire salamander S. corsica (four specimens montanus fe d mainly on benthic macroinvertebrates, in collected in the Ospedale region, and one from Haute particular Plecoptera, Ephemeroptera and Tricoptera Asco in Northern Corsica, all of them captured in ter­ larvae, while D. sardus primarily consumed terrestrial restrial habitats) was included for comparison, as this prey (Isopoda and Araneida). D. montalentii captured species feeds exclusively upon terrestrial invertebrates both freshwater, and terrestrial food items in similar (Kuzmin, 1994). The Corsican fire salamander sample proportions. These results suggest that the brook sala­ was used in the CA as a supplementary row element; mander was foraging mainly in deep water, while the thus, it was represented in the Euclidean space deter­ two discoglossids probably preferred surface or shal­ mined by the first two axes, but did not contribute to the low water. The frequencies of prey according to their calculation of the other species location (Greenacre, life-style (aquatic vs terrestrial) were overall signifi­ 1993 ). Correspondence analysis was performed using cantly different (x2=3 l.3, df=4, P<0.00 1), but when Minitab 11.12 statistical software (Minitab Inc., 1996). species were compared pairwise, the Corsican painted Prey volume frequency distributions were compared frogand the Corsican brook salamander showed similar with a Kolmogorov-Smimov two sample test (Siegel & feeding habits: D.montalentii - E. montanus, x2 =4.9, Castellan, 1988). The standardized version of Simp­ df=2; P=0.088; D. montalentii - D. sardus, x2 = 9.6, D (D )/(N D D. sardus - E. montanus, son's diversity index , = - 1 - 1 ), in which is df=2, P=0.008; x2=30.7, df=2, (D S N the Simpson's index = 11 P;2) and is the number P<0.001. The utilization of taxonomic categories by of prey taxa categories, was used to estimate trophic volume showed similar findings (Table 2), as D. niche breadth. This standardized index ranges from 0, montalentii and E. montanus were the species depend­ when only one resource category is exploited, to I ing the most upon aquatic food resources (58% and when all categories are exploited with the same fre­ 30% of the total ingested volume, respectively). The quency (Barbault, 1981 ). standardized Simpson index D, calculated on the basis of taxonomic categories, indicated that D. sardus had RESULTS the widest trophic niche breadth (D, = 0.769), E. During this study, 144 amphibians were stomach montanus the narrowest (D = 0.523), and D. , flushed(Table 1). The proportion of stomachs contain­ montalentii an intermediate value (D, = 0.544); but ing at least one prey item (inorganic and plant matter when prey volumes were considered, all species were excluded from data analyses) ranged from 100% showed similar - and relatively narrow values - ofniche in corsica to 63% in D. sardus, and a chi-square S. breadths (0.299 < D, < 0.356; see Table 2). analysis, based on the three largest samples (i.e. exclud­ The results of the CA based on the complete data set ing the S. corsica sample), indicated that there were no are shown in Fig. 1. The firstand second axes explained differences in the frequency of fasting animals (X2 = 38% and 24 % ofthe total variance, respectively. In this 4.4, df=2; P=0.1 08). plot, both Discoglossus sardus and S. corsica were pro­ The Euproctus montanus sample consisted of sexu­ jected in the bottom right quadrant of the plot, in strict ally mature individuals: 46 males and 6 females. The association with their terrestrial prey categories. sample ofD. sardus comprised 23 males, 6 females and Discoglossus montalentii and E. montanus were pro­ 2 subadults, and that of D. montalentii 2 males, 13 fe­ jected in different quadrants and their diets were males and 4 subadults. All subadult individuals were characterized by swimming invertebrates (Dytiscidae excluded from the quantitative data analysis. As the fre­ and Heteroptera) and by amphibian eggs or froglets, quencies of terrestrial and aquatic prey categories did repectively. not differ significantly between sexes [E. montanus The analysis of prey utilization by size showed that x2=0.2, df=l; D. sardus x2=3 .4, df=l; D. montalentii X2 in all species small food items (V<9 mm3) contributed a =1.8, df= 1; P>0.05 in all cases, all tests with Yate's cor­ large proportion (more than 50 %) of the total number rection forcontinuity (Siegel & Castellan, 1988)], food of prey, while relatively large items tended to be rather data from males and females were combined for infrequent. When prey-size distributions (Fig. 2) were interspecificcomparisons. compared, these amphibian populations showed similar DIET OF SYMPATRIC AMPHIBIANS IN CORSICA 165

TABLE I. Number of stomach-flushed specimens and mean snout-vent length (SVL) of adult D. montalentii, D. sardus, E. montanus and S. corsica containing at least one prey item in the stomach.

Species Sample size Sample with prey Adult sample with prey N N SVL (mm)±SD

D. montalentii 22 19 (86%) 44.4±9.57 (N = 15) D. sardus 49 31 (63%) 48.4±6.27 (N = 29) E. montanus 68 52 (76%) 49.6±3.69 (N = 52) S. corsica 5 5 (100%) 96.6±18.04 (N = 5)

TABLE 2. Food composition by taxa (N)and by volume ( V) expressed in mm3, and standardized trophic niche breadth (D) ofD. , montalenlii, D. sardus, E. montanus, and S. corsica.

D. montalentii D. sardus E. montanus S. corsica Prey type (15) (29) (52) (5) N(o/o) V(o/o) N(o/o) V(o/o) N(o/o) V(o/o) N(o/o) V(o/o) Aquatic I Coleoptera Dytiscidae (adults) 2(5) 149(13) 2 Coleoptera (adults)-Dytiscidae excluded 6(15) 47(4) 5(8) 28(2) 2(3) 5(0) 3 Coleoptera (larvae)-Dytiscidae excluded 4 (6) 64 (4) 3 (4) 10(0) 4 Trichoptera (larvae) 4(10) 367(32) 3(5) 116(7) 15(20) 328(13) 5 Ephemeroptera (larvae) 1(2) 25(2) 12(16) 170(7) 6 Plecoptera (larvae) 2(5) 13(1) 8(11) 28(1) 7 Heteroptera (adults) 3(7) 37(3) 2(3) 28(2) I (I) 15(1) 8 Heteroptera (larvae) 2(5) 14(1) 9 Vertebrata (eggs, larvae) 5(6) 192(8) Subtotal 20(50) 653(58) 14(22) 236(14) 46(61) 748(30)

Terrestrial 10 Isopoda 3(15) 68(4) 2(3) 16(1) II Myriapoda 2(5) 32(3) 6(9) 67(4) 2(17) 18(0) 12 Pseudoscorpionida I (I) 0(0) 13 Araneida 4(10) 35(3) 7(10) 73(4) 4(5) 11(0) 2(17) 13(0) 14 Gastropoda Pulmonata 1(2) 87(8) 6(9) 550(33) 2(3) 515(21) 6(50) 2694(99) 15 Collembola 2(3) 3 (0) 16 Diptera (adults) 1(2) 6(0) 5(8) 29 (2) I (1) 0(0) 17 Coleoptera (adults) 2(4) 9(1) 2(3) 24(1) 1(8) I (0) 18 Hymenoptera Formicidae 2(5) 16(1) 3(5) 12(0) 19 Hymenoptera Formicidae excluded I (2) 2(0) 20 Lepidoptera (larvae) 1(2) 14(0) Subtotal 12(30) 186(16) 35(54) 840(51) I 0(13) 542(22) 11(92) 2726(99)

Land, shallow water or mud-dwelling 21 Acarina I (I) 0(0) 22 Oligochaeta 2(5) 28(2) 6(9) 230(14) 4(5) 86(4) 23 Diptera (larvae) 1(2) 251 (22) 8(12) 334(20) 8(11) 858(35) I (8) 23(1) 24 Jnvertebrata unidentified 5(13) 22(2) 2(3) 14(0) 6(8) 9(0) 25 Vertebrata (Discoglossus froglets) I (I) 203(8) Subtotal 8(20) 301(26) 16(25) 578(35) 20(26) 1147(47) I (8) 23(1)

Total 40 1140 65 1654 76 2437 12 2749

Standardized niche breadth D, 0.544 0.300 0.769 0.299 0.523 0.356 166 S. SALVI DIO, R. SINDACO AND L. EMANUELI

, D.montalentii 8 ropean anurans (except forthe discoglossid Bombina), ' 1 9 which feed primarly on land, even during their aquatic breeding season (Chiminello & Generani, 1992; Nollert 1- & Nollert, 1995). When prey size distributions were analysed, there were no statistically significant differ­ 17 N 24 ences in prey utilisation between predators, despite 6 c 18 II! 4 13 their variation in body size (see Table 1). On the other c� 0 hand, important taxonomic prey differences appeared E.montanus 14 11 8 23 22 when foraging behaviours were analysed. Euproctus 25 9 16 21 15 montanus ingested mainly benthic hydrobionts, show­ 12 3 D.sardus ing a clear tendency to feed in deep water. This 10 -1 - foraging mode appeared qualitatively similar to that of S.corsica 20 E. asper D the congener Pyrenean brook salamander I I I (Montori, 1997). Conversely, D. sardus foragedmainly 0 1 -1 on land, while D. montalentii showed a mixed feeding Component 1 strategy capturing prey both in terrestrial and aquatic FIG. 1. Simple Correspondence Analysis plot showing the microhabitats in similar proportions. These findingsare association between the diets of sympatric D. montalentii, D. most likely due to increased time spent foraging in dif­ sardus, E. montanus and their food categories (numbers correspond to prey categories in TABLE 2). S corsica was ferent microhabitats by each species. Thus, these added as a supplementary row element (see text). observations suggest more aquatic foraging behaviour forthe Corsican painted frog, compared to the conge­ neric Tyrrhenian painted frog, at least during their reproductive period. % of prey 70 Prey type, and especially prey size, have been con­ sidered to play an important part in resource 60 partitioning in amphibian assemblages (Toft, 1985). Differences in foodsize were usually related to the • D. montalentii body size ofthe predator, while differences in prey type D D. sardus • E. montanus 40 were attributable to habitat preferences,or to the time in which food is available (Pilorge, 1982). In the present 30 case, amphibians were sampled in the same habitat, and thus the observed differences in food habits could be 20 related to species-specific trophic behaviour. On the

10 basis of these observations, these coexisting species seemed to partition food resources by adopting differ­

> ent foraging strategies, as contiguous microhabitats 0-9 20- 19 40-49 60-69 80-89 100 were exploited in different ways by the three species. FIG. 2. Prey size distributions in sympatric D. montalentii, ACKNOWLEDGEMENTS D. sardus and E. montanus from the Ospedale forest. This study was undertaken as a part of research patterns of food utilization (Kolmogorov-Smimov two fundedby the Ministere de !'Environnement Direction sample test, P>O. 10 for all comparisons). Regionale de !'Environnement de Corse (contract FN 237 53 95 OOO 14). Research permits (n° 96/263 and n° DISCUSSION 97/342) were obtained from the Direction de la Nature Discoglossus montalentii, D. sardus, and E. et des Paysages. We thank B. Recorbet and M. montanus were collected in or near aquatic habitats, Delaugerre for suggesting this research to us. G. where they were mating and spawning. At this time all Odiema performed chromosomal analysis on the three species should have been exposed to the same ar­ discoglossid frogs; S. Kuzmin, and P. Joly provided ray of prey items. This may not be true, for example, in usefulcomments on an early version of the manuscript. summer, when E. montanus adopts completely terres­ REFERENCES trial habits (Goux, 1953; Michelot, 1992) and D. montalentii is still active in or near aquatic habitats Barbault, R. (1981). Ecologie des Populations et des (Salvidio era/., unpublished data). The study species Peuplements. Paris: Masson. foraged on a wide variety of prey, suggesting a Boulenger, G. A. (1897-98). The Tailess Batrachians of generalist foraging strategy. The presence in their diet Europe. London: Ray Society. of freshwaterprey items indicated that they were all ca­ Chiminello, A. & Generani, M. (1992). Hy /a arborea L., pable of swallowing their foodunder water. Thus, these Pelobates fu scus insubricus C. and Rana esculenta findings confirm that Discoglossus are adapted to take "complex" feeding habits during the breeding period prey underwater (Boulenger, 1897-1 898) as newts do, a in some rice-fields of the Po Plane (Peimont, North rather unusual behaviour in comparison to all other Eu- Italy). In Sixth Ord. Meet. SEH, 115-120. Kors6s, Z. DIET OF SYMPATRIC AMPHIBIANS IN CORSICA 167

Z. and Kiss, I. (Eds). Budapest: Hungarian Natural Lanza, B., Vanni, S. & Brizzi, R. (I 992b). Discoglosse de History Museum. Corse Discogloss11s montalentii. In Atlas de Clarke, B. T. & Lanza, B. ( 1990). Note on the morphology Repartition des Batraciens et des Reptiles de Corse, and distribution of the Corsican Painted Frogs: 31-32. Delaugerre M. and Cheylan M. (Eds). Discoglossus sardus Tschudi and Discoglossus Pampelune, Spain: Pare Nature! Regional de Corse & manta/en/ii Lanza, Nascetti, Capula and Bullini. Boll. Ecole Pratique des Hautes Etudes. Mus. Reg. Sci. Nat., Torino, 8, 53 1-544. Michelot, M. ( 1992). L 'Euprocte de Corse Euproctus Delaugerre, M. & CheyIa n, M. ( 1992). Atlas de montanus (Savi, 1838). In Atlas de Repartition des Repartition des Batraciens et des Reptiles de Corse. Batraciens et des Reptiles de Corse, 21-23. Pampelune, Spain: Pare Nature! Regional de Corse & Delaguerre, M., and Cheylan M. (Eds). Pampelune, Ecole Pratique des Hautes Etudes. Spain: Pare Nature! Regional de Corse & Ecole Fraser, D. F., ( 1976). Coexistence of salamanders in the Pratique des Hautes Etudes. genus Plethodon: a variation of the Santa Rosalia Minitab Inc. ( 1996). Minitab Reference Manual. Released theme. Ecology 57, 238-25 1. 11 for Windows. Printed in the USA. Gase, J.-P., Cabela, A., Crnobrja-lsailovic, J., Dolmen, Montori, A. ( 1997). Trophic segregation between the D., Grossenbacher, K., Haffner, P., Lescure, J., Pyrenean newt (Euproctus asper) and the brown trout Martens, H., Martinez Rica, J. P., Maurin, H., (Sa/ma /rut/a fa rio). In /-/erpetologia Bonnensis. Oliveira, M. E., Sofianidou, T. S., Veith, M., & Bohme, W., Bischoff, W. and Ziegler, T. (Eds). 273- Zuiderwijk, A. (1997). Atlas of Amphibians and 278. Bonn: SEl-1. Reptiles in Europe. Paris: Socictas Europaea Noller!, A. & Nollert, C. ( 1995). Los Anfibios de Europa. Herpetologica & Museum National d'Histoire ldentificaci6n, Amenazas, Protecci6n. Barcelona: Naturelle (IEGB/SPN). Ediciones Omega, S.A. Goux, L. (1953). Contribution a l'etude biogeographique, Pilorge, T. ( 1982). Regime alimentaire de Lacer/a ecologique et biologique de I 'Euprocte de Corse vivipara et Rana temporaria dans deux populations Euproctus montanus (Savi) (Salamandridae). Vie sympatriques du Puy-de-D6me. Amphibia-Reptilia 3, Milieux 4, 1-36. 27-3 1. Green acre, M.J. ( 1993). Correspondence Analysis in Sansoni, G. ( 1988). A tlante per ii riconoscimento dei Practice. New York: Academic Press. macroinvertebrati dei corsi d 'acqua italiani. Trento, Joly, P. (1987). Le regime alimentaire des Amphibiens: Italia: Provincia Autonoma di Trento. methodes d'etude. Alytes 6, 11-17. Siegel, S. & Castellan, N. J. ( 1988). Non parametric Kuzmin S. ( 1994 ). Feeding ecology of Salamandra and statistics for the behavioural sciences. Singapore: Mertensiella: a review of data and ontogenetic McGraw-Hill International Editions. evolutionary trends. Mertensiella 4, 271 -286. Toft, C.A. ( 1985). Resource partitioning in amphibians Lanza, B., Nascetti, G., Capula, M. & Bullini, L. (1984). and reptiles. Copeia 1985, 1-21. Genetic relationships among West Mediterranean Discoglossus with the description of a new species (Amphibia, Salientia, Discoglossidae). Monitore zoo/. ital., (N.S.) 18, 133-152. Lanza, B., Vanni, S. & Brizzi, R. (I 992a). Discoglosse de Sardaigne Discoglossus sardus. In Atlas de Repartition des Batraciens et des Reptiles de Corse, 29-30. Delaugerre M. and Cheylan M. (Eds). Pampelune, Spain: Pare Nature! Regional de Corse & Ecole Pratique des Hautes Eludes. Accepted: 5. 5. 99